Many of the prevailing features in networks today have their roots in technologies developed for the existing Internet, rather than the anticipated evolution of future complex networks. The more recent concept of network coding goes beyond traditional forwarding of immutable packets, viewing network operation instead as flows of information that can be operated on mathematically within the network. In particular, network nodes may perform mathematical operations across packets received on dierent incoming links to form new packets containing coded combinations of the original packets' information, which will be subsequently decoded at other nodes in the network. Traditional store-and-forward operation, where network nodes simply route packets unchanged from incoming to outgoing links, is a special case of the more general and powerful network coding framework. Advantages of network coding have been shown, in theoretical analysis as well as experimental implementations, for various aspects of network operation, such as improving throughput, increasing loss resilience and reducing network control overhead. Network coding is particularly promising for resource-constrained and uncertain network scenarios.
Although the theoretical foundations of network coding are well understood, real-world systems need to solve numerous practical aspects before network coding meets its promised potential. As a result, network coding systems need to make numerous practical design choices and optimizations that are essential to leverage network coding and achieve good performance. Unfortunately, such designs did not consider security aspects, and consequently many of these choices resulted in protocols that have numerous security vulnerabilities. For example, the very nature of packet mixing makes network coding systems vulnerable to a severe security threat known as pollution attacks, in which attackers inject corrupted packets into the network. This attack has a devastating impact on network coding because it results in an epidemic propagation of corrupted packets, as further nodes code and forward more corrupted packets. However, packet pollution is only one of many potential attacks. For example, the complexity of network coding systems, as well as the inherent vulnerability of multi-hop wireless networks, create numerous opportunities for attacks against network coding systems for wireless networks.
Next: Call for Participation
Next: Program
Workshop Index
DIMACS Homepage